search

hexylena / docker-rstudio-notebook

RStudio as a Galaxy Interactive Environment
MIT License
5 stars 17 forks source link

Should we add miniconda3 to the image? #28

Closed bgruening closed 2 years ago

bgruening commented 5 years ago

This would make it very easy to install packages for users and would avoid domain-centric images.

yvanlebras commented 5 years ago

Ca nbe of interest notably if you consider that sometimes, trying installing R packages from R Studio GUI is asking you to install OS libs from a terminal like here:

** package ‘xslt’ successfully unpacked and MD5 sums checked
** using staged installation
Using PKG_CFLAGS=-I/usr/include/libxml2
Using PKG_LIBS=-lexslt -lxslt -lxml2
------------------------- ANTICONF ERROR ---------------------------
Configuration failed because libexslt was not found. Try installing:
 * deb: libxslt1-dev (Debian, Ubuntu, etc)
 * rpm: libxslt-devel (Fedora, CentOS, RHEL)
 * csw: libxslt_dev (Solaris)
If libexslt is already installed, check that 'pkg-config' is in your
PATH and PKG_CONFIG_PATH contains a libexslt.pc file. If pkg-config
is unavailable you can set INCLUDE_DIR and LIB_DIR manually via:
R CMD INSTALL --configure-vars='INCLUDE_DIR=... LIB_DIR=...'
--------------------------------------------------------------------
ERROR: configuration failed for package ‘xslt’
* removing ‘/usr/local/lib/R/site-library/xslt’
Warning in install.packages :
  installation of package ‘xslt’ had non-zero exit status
hexylena commented 5 years ago

Does that mean all R packages will need to be installed through conda? That feels kinda weird since R has its own solutions to these issues (namely MRAN's "time machine")?

yvanlebras commented 5 years ago

hmmm really interesting, true. But ITM, MRAN forces to use Microsoft corporation R distrib... I never test it but if this can be of interest, maybe it's time for me to look at it.

bgruening commented 5 years ago

Does that mean all R packages will need to be installed through conda?

yes

That feels kinda weird since R has its own solutions to these issues (namely MRAN's "time machine")?

Which has its own limitations. And I expect many people need to have system dependencies either way. netcdf, boost, or even samtools are probably a valid use-case. You get all those from conda and we can fix problems easily if they arise which we can not with MRAN.

hexylena commented 5 years ago

If someone has motivation for implementing it, then, yeah, ok, could make sense.

Op di 15 okt. 2019 14:22 schreef Björn Grüning notifications@github.com:

Does that mean all R packages will need to be installed through conda?

yes

That feels kinda weird since R has its own solutions to these issues (namely MRAN's "time machine")?

Which has its own limitations. And I expect many people need to have system dependencies either way. netcdf, boost, or even samtools are probably a valid use-case. You get all those from conda and we can fix problems easily if they arise which we can not with MRAN.

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/erasche/docker-rstudio-notebook/issues/28?email_source=notifications&email_token=AADP7O3LIVB4JDTAGB75XKTQOWYZNA5CNFSM4JAPZYZKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEBIRKOA#issuecomment-542184760, or unsubscribe https://github.com/notifications/unsubscribe-auth/AADP7O6E2D65R7DYYIOGYSLQOWYZNANCNFSM4JAPZYZA .